# Load model directly
from transformers import AutoTokenizer, AutoModelForMaskedLM
tokenizer = AutoTokenizer.from_pretrained("Billwzl/20split_dataset_version3")
model = AutoModelForMaskedLM.from_pretrained("Billwzl/20split_dataset_version3", device_map="auto")Quick Links
20split_dataset_version3
This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.8310
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 16
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.1679 | 1.0 | 313 | 2.9768 |
| 2.9869 | 2.0 | 626 | 2.9299 |
| 2.8528 | 3.0 | 939 | 2.9176 |
| 2.7435 | 4.0 | 1252 | 2.9104 |
| 2.6458 | 5.0 | 1565 | 2.8863 |
| 2.5865 | 6.0 | 1878 | 2.8669 |
| 2.5218 | 7.0 | 2191 | 2.8802 |
| 2.4647 | 8.0 | 2504 | 2.8639 |
| 2.3933 | 9.0 | 2817 | 2.8543 |
| 2.3687 | 10.0 | 3130 | 2.8573 |
| 2.3221 | 11.0 | 3443 | 2.8398 |
| 2.276 | 12.0 | 3756 | 2.8415 |
| 2.2379 | 13.0 | 4069 | 2.8471 |
| 2.2427 | 14.0 | 4382 | 2.8318 |
| 2.1741 | 15.0 | 4695 | 2.8356 |
| 2.1652 | 16.0 | 5008 | 2.8310 |
Framework versions
- Transformers 4.21.0
- Pytorch 1.12.0+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1
- Downloads last month
- 4
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="Billwzl/20split_dataset_version3")